Problem 3
Inverting the polygon at one of its own vertices turns the cyclic polygon into a set of collinear image points, and Ptolemy's inequality on the original circle becomes an exact linear relation among the inverted distances.
Suppose for contradiction the statement fails for some -gon (all sides' squared lengths divisible by ) with minimal among counterexamples, so every diagonal fails to be divisible by in the same squared sense (otherwise a diagonal would split the polygon into two smaller cyclic polygons with all relevant squared lengths divisible by , to which the inductive hypothesis on smaller would apply). Write and invert the plane at with radius . Since inversion sends the circumcircle of the to a line, the images of become collinear, and the identity (a telescoping sum of inverted-distance segments along that line) is exactly a generalized Ptolemy relation for the inverted points.